Even in today’s age of film and TV streaming, automated DVD rental kiosks, and advances in home theater systems, yearly box office earnings are continuing to increase. In fact, they are higher than ever before.1 People are still heading out to cinemas in droves, looking for that unique and visceral experience where they can lose themselves in the larger-than-life visuals and sounds.2 Dolby Laboratories, which pioneered surround sound in film with the original Star Wars in 1977, has remained a leader in creating technologies for the entertainment industry for 50 years — advancing the science of sight and sound to enable spectacular, immersive, emotionally charged experiences for all movie-goers and entertainment enthusiasts.
“Dolby is a company that is involved in the whole ecosystem of the industry, and our customers have a very deep partnership with us,” says David McDonough, Director of BI Analytics at Dolby Laboratories. “We work with movie directors, sound mixers, and editors to make the most solid movie imaging possible and then also provide cinemas with the different servers, technology, and equipment required to play the movie with the best viewing experience.”
This Dolby technology is then offered for multiple consumer devices — that is TVs, mobile phones, laptops, digital set top (cable) boxes, or anything else that can play back sound or image. To consistently deliver on its mission to ensure all customer experiences are the best they can be, the business itself also must rely on technology; Dolby completed a nine-month SAP implementation back in October 2011. That go-live included SAP ERP 6.0, SAP Business Warehouse (SAP BW) 7.0, SAP BusinessObjects Planning and Consolidation 7.5, SAP BusinessObjects solutions, and the SAP Incentive Administration application by Vistex. The solution suite was deployed concurrently and rolled out as one global instance. Later, in September 2015, Dolby decided to upgrade to SAP BW 7.4 and SAP BusinessObjects Planning and Consolidation 10.1 in anticipation that a move to SAP HANA was imminent.
Dolby has roughly 500 business intelligence (BI) users worldwide who yield around 11,000 BI application uses each quarter. McDonough, who oversees all enterprise forecasting and BI in his role at Dolby, is tasked with bringing data sets together through the organization’s key business applications (including SAP and non-SAP systems), mashing up that data, and transforming it into actionable information so the business can make better-informed decisions, close the books more accurately, and ensure optimal service levels and support to its customers and partners. “We need a complete picture of our customers to fully understand them, and a lot of data is generated between us that is very relevant — such as contractual or royalty statement information — that we need to bring together to create a picture of the entire licensing lifecycle of their current Dolby technologies and what next generation of mobile phone or TV they may port to the next Dolby technology,” he says. “When Dolby employees are interacting with customers, we should be able to bring up complete storyboards right away showing their past history, current information, and future path so we can advance the customer relationships and the possible opportunities for them.”
Everything operational in the company — from forecasting to tracking actuals, closing deals, closing sales at quarter end, and closing the books — is driven off the enterprise data warehouse. Adding volume to the data warehouse is the fact that Dolby is emerging in new territories and launching new products into the marketplace fairly quickly. With the company expanding, Dolby must be able to drive business information to identify areas of projected growth. “We want to anticipate business needs by providing a deeper and richer data set across particular spaces, whether it’s the whole sales operation cycle, the complete customer cycle, or a new product initiative,” McDonough says.
The BI group came up with some strategic analytics imperatives — such as driving business information to advance corporate strategies forward, making BI easier to use, increasing adoption, and consolidating reporting — and discovered that there were some clear challenges in its current landscape.
The most apparent challenge of the existing BI environment was performance. Reporting was optimized under the hood as much as possible, but without changing the engine, it was slow from most end-user perspectives. “We couldn’t build storyboards as broadly as we wanted because we were limited by the number of cubes we could build and reconnect without getting massive data response issues,” says McDonough. “We also spent significant time reworking existing data models to build cubes to address business needs.”
Data redundancy was also creating headaches. Ultimately, building data cubes involved splittingup data to run faster in specialized data cubes, but that often resulted in redundancy. In some cases, there could be as many as three levels of the same data in the SAP BW system, which significantly increased the size of the database and the effort required to maintain it.
The high complexity and layering of the data created a lag. “We typically refreshed data nightly, running a batch process that took about six hours to monitor and check,” says McDonough. “Some refreshes did run every hour, but we had limited capability for fast refreshes. And during those six hours, our EMEA BI users often experienced slow performance because of these lagging data refreshes.”
The final challenge was the lack of agility. With such a complex data cube chain to maintain, mashing up many-to-many or one-to-many data sets was taking a toll on the data performance and architecture. “We maximized capabilities and got great use out of the current database, but we started to hit the threshold of what was possible. We outgrew our crib, in a sense, and now we need a bigger bed,” says McDonough. “Going to the next-generation database would put us into a huge ocean of possibilities where we could go in any direction we want.”
For these reasons, the business resolved that the next logical step was to enhance its enterprise data warehouse and adopt an in-memory database.
We outgrew our crib, in a sense, and now we need a bigger bed. Going to the next-generation database would put us into a huge ocean of possibilities where we could go in any direction we want.
—David McDonough, Director of BI Analytics, Dolby Laboratories
Making a Resounding Case
At first Dolby wasn’t sure that SAP HANA would be the right solution to fit the company’s strategic need. “We aren’t a traditional big data company where we constantly refresh massive amounts of data throughout the day that we have to crunch in real time,” says McDonough. “But we wanted a powerful database to bring in data sets from all over, mash up the information, and provide it to users quickly as interesting storyboards — and do this without killing performance but rather improving it.”
As a first step, the BI team created a secondary relational database with a subset of expansive key business data sets from various sources. This parallel data store ran fast and was fairly easy to build out, and pushed the needle on performance, making end users happier. “The project gave us a sense of confidence, and it gave the business a sense of the capability we can provide with the proper system,” says McDonough. “The experience helped us go to the next step to evaluate in-memory databases that fit best for Dolby.”
After reviewing the in-memory offerings in the market, Dolby was hard-pressed to find anything more robust than what SAP HANA brought to the table. The solution exceeded expectations for the data agility and integration capabilities the business was looking for and had a strong roadmap that aligned with the company’s future strategy. A second advantage Dolby discovered was the runtime SAP BW powered by SAP HANA license, which the business found to be much more attractive than the full license option. Two caveats for the runtime license were 1) Dolby could only run SAP HANA in the SAP BW environment, and 2) all the presentation tools had to be SAP BusinessObjects solutions. But because Dolby was using those solutions already, those caveats became advantages. It was a great fit, according to McDonough, and a nice upside was that the business could load as much data as it wanted into that environment. “Whether we had a 1TB or a 20TB system, we would pay the same price; this model is attractive for companies looking to take their BW environments to the next level, where SAP BW is the operational and analytics engine for the business,” he says. “While many companies have more decentralized landscapes, we have matured our BI footprint and brought in very rich data sets to our SAP BW environment. We can still bring in other non-SAP data sources — such as data from our hosted customer relationship management application, our legal contracts application, and human resources systems — under the runtime license as long as it goes through SAP BW, which works out great because we have a centralized data warehouse and want to do all our transformations in one place.” So with SAP HANA looking like the clear winner, the decision was made to implement SAP BW powered by SAP HANA as the next-generation data warehouse.
Putting the Picture Together
Next, the BI team put together a pricing model and a five-year roadmap, brought it to senior leadership, and got the green light to kick off the project. The project was set to begin in February 2016 and was scheduled to take nine months, concluding in early November 2016. The original plan was a migration of data from SAP BW 7.4 to SAP HANA; however, during the evaluation phase, SAP released SAP BW 7.5. So the project scope was increased to an upgrade and a migration. “We reviewed the new release and decided to take advantage of the new enhanced functionality. SAP provided a script that performed both the upgrade to 7.5 and the migration from the existing database to SAP HANA concurrently, all executed flawlessly and with minimal risk.”
From the outset, the project was intended to be a technical upgrade rather than functional and to have everything work out of the box with little-to-no customization. During the course of the upgrade, the team applied only a handful of service notes and went through a couple service pack updates to get up to SAP BW 7.5.3 powered by SAP HANA 12, which is currently the latest version.
From an environment standpoint, Dolby has a classic SAP BW landscape including development, quality assurance, and production servers, which are all virtual servers. The first four months of the project plan were all allotted to performance evaluation and regression testing a copy of production in a virtual sandbox. The sandbox upgrade and migration to SAP BW 7.5 powered by SAP HANA was performed in one week by a project team that included partner resources. “We collaborated with a private cloud/hosting partner on the design and approach, they did the hands-on migration, upgrade, and Basis work, and then we performed the post-regression testing and some of the configuration,” says McDonough. “The migration and upgrade will only take a weekend for production. We spent half of the project time with the sandbox to ensure we got everything right so that during the other half, which kicked off in July 2016, we can move quickly through development, quality assurance, and production.”
Once the sandbox upgrade was complete, attention turned to database sizing and testing. The expectation prior to the migration was that Dolby would need a 1TB SAP HANA server — under current standards, anything above a half a terabyte has to be a physical server. After the migration, the team redid the SAP HANA sizing exercise and recognized it had a much higher compression ratio, which is a testament to the team’s efficient data structure and pre-migration modeling designs. The outcome of that post-migration sizing determined Dolby would only require a half terabyte server, which meant it could be a private virtual server, giving the business more flexibility and also less of a hosting expense.
The team then spent a few weeks trying out the new functionality, performing complete regression and methodical performance testing on a variety of heavy-duty queries to see how the SAP HANA engine would perform at the San Francisco headquarters and at office locations outside of the Americas. End-user performance was also tested rigorously.
The project so far has been without any real challenges or setbacks, according to McDonough, who attributes that in large part to the team’s heavy planning efforts and focus on the future. “We are challenging ourselves to think about how and where we want to approach our new future of SAP HANA views,” he says. “The ABAP coding transformations we already developed in the past will stay because they work, and it doesn’t make sense to rebuild them. But we will strip away all the cubes and shrink down the aggregates so there is less to maintain. Then new future development will be written off the data objects that feed directly to SAP HANA views. It’s as if SAP BW becomes a shell, and all the real work is done in SAP HANA. Then, we just connect the old world of our existing structures with the new world where relevant.”
Achieving Optimal Performance
As far as results go, the business should expect to see fundamentally faster performance across the whole enterprise. “People will be getting the information they want immediately instead of waiting for answers, so they are more invested in drilling down for deeper insights,” McDonough says. “There is nothing worse than wasting time watching a little circle spin, and with this level of performance, all that nearly goes away. It was amazing to see how we could push massive amounts of data around and build cool storyboards in seconds.” For example, a complex BI summary reports that used to take 4.5 minutes now is delivered in five seconds, which is 54 times faster.
Complex queries across the different SAP BusinessObjects solutions through SAP BW powered by SAP HANA showed a 10 times performance improvement on average. That means end users will have much more time to devote to BI, analytics, and other refinements. McDonough projects that performance gains on pure SAP HANA — without data passing through the presentation tools — would be even higher than that. SAP BusinessObjects Planning and Consolidation, which has a lot more calculations and script logic involved, also showed improved performance. When input templates (where data is entered) were put into SAP BW powered by SAP HANA, performance was calculated at three times faster and reports were two or three times faster, depending on which one was run. Finally, the overall landscape footprint was reduced by 85%, compressing 1.5TB of data down to 130GB.
McDonough describes the project as seamless and one of the best experiences he has had with an SAP upgrade or migration. “Now, we can consume data any way we want, create faster storyboards, and provide quicker insights, which are the real value-add we were after. More than just an in-memory database, SAP HANA is a mature platform with lots of functionality — and SAP keeps adding more and more to it. So when I look at where SAP is going with this platform, I feel it is a very sound investment for Dolby for a very bright, scalable future.”
1 In 2015, 1.3 billion movie tickets were sold in the US alone, totaling $11.3 billion in box office earnings: the-numbers.com/market. [back]
2 Star Wars Episode VII: The Force Awakens, which broke countless records, grossed over $2 billion in box office sales worldwide in 2015, reaching $1 billion in just 12 days: boxofficemojo.com/movies/?id=starwars7.htm. [back]